COMPLIANCE & RESPONSIBLE AI

Built audit-ready. By design.

Most AI hiring tools were built before the laws caught up. Taskflow Recruit was built after — with NYC Local Law 144 disclosure on every call, EEOC-aligned scoring, adverse-impact monitoring on every cohort, and an independent bias audit each quarter.

NYC LL 144 · EEOC · ADEA · SOC 2 TYPE II (IN PROGRESS) · GDPR · CCPA · HIPAA-AWARE
OUR POSITION

We score answers. You make decisions.

Taskflow Recruit's AI interviewer asks questions and scores answers against the rubric you wrote. It does not decide who to hire. It does not rank candidates without your sign-off. It does not learn from your past hiring decisions — because that's where bias enters. Every score is traceable to a question, an answer, and a rubric line. Every cohort is monitored for adverse impact. Every candidate is told they're talking to AI before the interview starts.

This is what "responsible AI in hiring" actually looks like.

NEW YORK CITY LOCAL LAW 144

NYC LL 144, out of the box.No checkbox. No add-on. No "coming soon."

Automated Employment Decision Tool (AEDT) disclosure

Every candidate hears a clear AI disclosure at the start of every call: "This is an AI conversation. Your responses will be reviewed by a recruiter. You may request a human interview at any time."

CLAUSE 20-870(2)
Annual bias audit (independent, public)

We commission an independent bias audit annually from a third-party auditor. The most recent summary is published on this page. We never grade our own homework.

CLAUSE 20-871(a)
10-day candidate notice

Candidates in NYC receive notice at least 10 business days before any AI screening, with instructions for requesting alternative selection processes. We handle this via SMS automatically when ATS detects a NY address.

CLAUSE 20-871(b)
Selection-rate reporting

Every hiring cycle, we publish selection rates by race, ethnicity, and sex to your compliance dashboard. Export to PDF for retention or external audit.

CLAUSE 20-871(c)
Plain-language explanation

Every score includes a plain-language rationale tied to specific rubric criteria. Candidates can request a written explanation of why they advanced or didn't.

RIGHT-TO-EXPLANATION
Data retention policy

All interview audio, transcripts, and scores deleted at the candidate's request within 30 days. Default retention is 12 months unless your contract specifies otherwise. We never train models on customer data.

DATA HANDLING
LATEST BIAS AUDIT: Q1 2026 · AUDITOR: [PLACEHOLDER — INDEPENDENT FIRM] · [DOWNLOAD SUMMARY →]
FEDERAL ALIGNMENT

Aligned with EEOC, ADEA, ADA, and Title VII guidance.

The EEOC's 2023 technical assistance document on AI in hiring made one thing clear: employers are responsible for adverse-impact testing on any algorithmic screening tool they use. Vendors who tell you "we're EEOC compliant" without showing you the math are lying.

Taskflow Recruit runs the 4/5ths rule on every cohort automatically. If your hiring is selecting one protected group at less than 80% the rate of the highest-selected group, you'll see a flag in your dashboard before you make an offer. We give you the data. You decide what to do with it.

We also block protected-class questions in real time. If a candidate volunteers protected information, the AI does not include it in the scorecard, does not transcribe it for the recruiter, and flags the moment for audit.

BLOCKED IN REAL TIME
  • Age
  • Marital status
  • Race
  • Pregnancy
  • Religion
  • Disability status
  • National origin
  • Sexual orientation
  • Gender identity
  • Family medical history
  • Citizenship
  • Genetic information
12 PROTECTED CATEGORIES · UPDATED QUARTERLY · CONFIGURABLE PER JURISDICTION
THE WORKDAY PROBLEM

Workday is still defending its old tools in court.We were built after the ruling, not before it.

In February 2023, Derek Mobley filed a class-action lawsuit alleging Workday's AI hiring tools systematically discriminated against applicants over 40, applicants with disabilities, and Black applicants. In May 2025, Judge Rita Lin of the U.S. District Court for the Northern District of California granted preliminary collective certification — meaning anyone in the U.S. who was screened by Workday's AI tools between 2018 and now may now opt into the class.

Thousands of companies use Workday's hiring tools. Many will be named.

Taskflow Recruit doesn't have that exposure. Our AI does not make hiring decisions — it scores answers against the rubric you write. It doesn't learn from your historical hiring data. It blocks protected-class topics in real time. It hands the recruiter a defensible audit trail for every decision.

We were built after the case, with the case in mind. Compliance is not a feature we added. It's the architecture.

Citations to Mobley v. Workday are factual references to publicly filed court records. Allegations described in those filings have not been adjudicated on the merits. Taskflow makes no claim about Workday's product beyond what is publicly described in court filings.

SECURITY & DATA

Where your data lives, how long it stays, who can see it.The boring details that matter.

Encryption in transit and at rest

All interview audio and transcripts encrypted with AES-256 at rest, TLS 1.3 in transit. Infrastructure hosted on AWS (us-east-1 primary, us-west-2 failover). SOC 2 Type II audit in progress, expected Q3 2026.

RBAC, SSO, audit logs

Role-based access control with seven default roles. SAML SSO supported via Okta, Azure AD, Google Workspace, OneLogin. Every action by every user is logged with an immutable audit trail. Export to SIEM.

Candidate-controlled retention

Candidates can request deletion of their interview audio, transcript, and scorecard at any time. Default retention is 12 months. Per-customer contracts can specify shorter retention. We never train models on customer data. Period.

INSIDE THE PRODUCT

Your compliance dashboard. Always live.

SELECTION RATE
84%
+2 pts vs Q1
ADVERSE IMPACT
✓ PASSING
4/5ths rule met
NYC LL 144 NOTICES SENT
342 / 342
this cycle
SELECTION RATE BY GROUP
Asian
10%
Black
10%
Hispanic
10%
White
10%
Female
10%
Male
10%
Age 40+
10%
Age <40
10%
✓ ALL GROUPS WITHIN 4/5THS RULE
AUDIT LOG
2026-05-12 14:08:11LL144 notice sentCandidate #4827NYC
2026-05-12 14:07:53Protected-class flagblocked, redactedAUTO
2026-05-12 14:02:11Adverse-impact recheckPASSQ2 cohort
2026-05-12 11:43:09Bias audit Q1 publishedsummary availablePUBLIC
2026-05-12 09:21:44Candidate deletion requestfulfilledUSER
DASHBOARD AVAILABLE ON GROWTH, SCALE, AND ENTERPRISE TIERS · LIVE DEMO ON REQUEST
FREQUENTLY ASKED

What legal teams ask us most.

"EEOC compliant" isn't a thing a vendor can be — compliance is a property of the way an employer uses a tool, not the tool itself. What we can say is that we're aligned with the EEOC's 2023 technical assistance on AI in hiring: we run the 4/5ths adverse-impact rule on every cohort, we block protected-class topics in real time, and we make every score traceable to a rubric criterion. We give your legal team what they need to defend a disparate-impact claim. Your legal team makes the final call.

Stop reading.Get interviewed.

The fastest way to understand what we do is to be on the other side of the call.